3.1. Experiment Design and Set-Up
In the experiments, a DMG HSC 75 linear high-speed machine was used to acquire the higher cutting speed. In order to reduce the random error occurring in the experiments, all the experiments were done twice. The milling cutter’s diameter was 12 mm, with a six-flute monolithic ceramic milling tool. SiAlON was the material used for the milling tool, and its base material was Si3N4. The tool was manufactured by Kennametal Inc.
As for the experiment parameters, the experiments only changed the spindle speed, which was proportional to the cutting speed. As the cutting speed was in direct proportion to cutting temperature at the shear zone, the change of cutting speed could lead to a change of milling temperature in the shear zone. In this experiment, milling speed was set to varying from 370 m/min to 621 m/min. (Because the highest spindle speed of the machining center was 18,000 rpm, for safety reason, spindle speed was limited to 16,500 rpm. For the 12 mm ceramic tool, cutting speed was limited to 621 m/min). The cutting depth was set to be 0.3 mm, and the cutting width was set to be 5 mm. Because the parameters used in the experiment frequency was lower than the cutting process natural frequency, there was no resonance during the experiment process. Thermal maps were captured by a FLIR T640 45° infrared camera (FLIR, Portland, Oregon, USA). Thermal maps were captured by a FLIR T640 45° infrared camera. The infrared camera could capture thermal temperature ranges from +300 °C to +2000 °C. The infrared camera was mounted on a tripod. The milling force was detected through a Kistler 9129 AA dynamometer (Kistler Group, Winterthur, Switzerland) mounted on a self-designed jig. The experiment setup is shown in
Figure 8. The dry milling process is shown in
Figure 9 (because the workpiece material was a forging blank, shown as black on its side surface and before the milling experiment, the forging blank on the up and down surface was cleaned to avoid measurement error in the experiments).
3.2. Experiment Results and Discussion
The cutting force was obtained from the experiments, and the average cutting force was calculated by Equation (12).
As
Figure 10 shows, with the rise of spindle speed, the milling force first remained constant until milling speed reached 527.52 m/min (Spindle speed 14,000 rpm), which was a turning point for milling force, after when the force decreased sharply. The behavior of the simulated milling force and the real milling force acquired from the force dynamometer was the same.
Because the experiments were conducted by changing the spindle speeds for the purpose of changing the cutting speed, the experiments could not get an integer number for the cutting speed. The behavior of the milling force in both experiments and simulation was compared. The deviation between the actual milling force and random milling force was 5.37%.
The captured milling temperature was analyzed by focusing on the square area of the shear zone, as shown in
Figure 11. The square area in the thermal map was roughly focusing on the shear zone of the dry milling process. The average temperature in the square area was captured and then analyzed by software, from which a brief understanding of the dry milling temperature in the experiments could be obtained. The FLIR T640 45° thermal camera had an accuracy of ±2 °C, and the accuracy was greatly guaranteed by the thermal camera.
The values measured by the infrared camera were based on the Stefan–Boltzmann law proposed by Cercignani [
30], which indicates that the total radiant heat power emitted from a surface is proportional to the fourth power of its absolute temperature.
Equation (13) is the formula of the Stefan–Boltzmann law. In this equation,
T represents absolute temperature, and
ε refers to the emissivity. Because the milling temperature was captured through a small square in the cutting region, the emissivity could be considered as a constant. For Inconel 718, when the temperature was 20 °C, Inconel 718′s
ε was taken to be 0.25. This was set in the thermal camera post-processing software. The software could adjust it automatically when the temperature was changing.
σB is the Stefan–Boltzmann constant,
σB = 5.67 × 10
−8 W/(m
2·K
4).
In this research, the infrared-camera-captured initial temperature was based on the
ε value of 1. This primitive temperature was converted by changing
ε to 0.25. Thus, the actual average temperature value was obtained, and the comparison between the actual value and simulation value with the change of cutting speed is shown in
Figure 12, and the thermal image of the highest cutting speed is shown in
Figure 13. It is obvious that the average temperature in the simulation was higher than the average temperature in the experiments. The major reason for this phenomenon is that the temperature acquired in the real milling process could not reflect the shear zone’s highest temperature, and the heat dissipation in the environment could lead to temperature value lower than the simulation temperature.
According to the experiments, the tendency of average temperature was to steadily increase with the increase of cutting speed, until when the spindle speed reached 14,000 rpm (cutting speed 527.52 m/min), and the average temperature of cutting region was 653.8 °C. Then, with the improvement of spindle speed, the average temperature in the cutting region exceeded 800 °C, which already exceeded the Inconel alloy material softening point. As shown in
Figure 10, when spindle speed reached 14,000 rpm (cutting speed is 527.52 m/min), the cutting force began to decrease and continued to decrease with increasing cutting speed.
Given the close relationship between the cutting temperature in the shear zone and the cutting force generated in the machining process, by taking the cutting temperature as the horizontal axis value and the cutting force as the vertical axis value, the scatter plot was drawn, as shown in
Figure 14. The two-term Gaussian curve could interpolate the observations. The interpolation curves of Gaussian, the cutting force, and the cutting temperature were obtained. The two-term Gaussian curve was the most precise interpolation curve for the force and temperature. It could clearly reflect the relationship between these two values. The confidence bounds were 95%.
The curve could be expressed by Equation (14). Wherein, t represents the average temperature value.
According to the curve, it can be concluded that when the average milling temperature was at 800 °C, the average cutting force reached a turning point. It indicated when the cutting speed was at 560 m/min, there was a material softening phenomenon on the Inconel 718, which led to the cutting force reduction.
The initial increase of the milling speed induced the average temperature to increase, and the interface between tool and workpiece material might melt and lead to the force reduction. As the temperature continued to increase, more material began to melt and tended to cut more materials, which made the milling depth to increase, as well as increase the milling force. Although the force value fluctuated before the material softening effect, the average milling force at the initial stage almost kept a constant value.
As
Figure 14 shows, before the material softening effect happened, the milling force had a relatively constant value. Through calculating the average force value and considering the material softening effect rate, the average milling force before the material softening effect was shown to be 589 N. The force reduction rate (
r) was calculated as Equation (15), which is
f (
t) minus the average force and then divided by the average force. Because it was a division of two force values, so its unit was 1.
It can be concluded in
Figure 15 that when the temperature was at 200 °C to 800 °C, the cutting material experienced a work hardening effect. The force reduction started at 800 °C and continuously increased with the temperature thereafter. It can be concluded that when the cutting speed reached a threshold, the cutting heat generated at the machining region was large enough to cause the material softening effect for the nickel-based superalloy.
3.3. Tool Wear, Microstructure of Machined Surface, and the Form of Chips Under Different Cutting Parameters
The used ceramic tool wear was observed through an SEM (scanning electron microscope, Zeiss, Aallen, Germany), as shown in
Figure 16 and
Figure 17. The main wears on the ceramic tool when the cutting speed was above 600 m/min were the material bonding and diffusion wear on the tool surface. When the cutting speed was below 600 m/min, especially when the cutting speed was at cutting force turning point, the main tool wear was tipping, as shown in
Figure 18.
When considering the cutting length-induced tool wear, a tool life graph could be drawn, as shown in
Figure 19. When cutting length reached 0.6 km, the tool back surface wear began to wear severely.
After the dry milling process, the cut material was analyzed through a metallographic microscope. The dry milling process used nickel-based alloy and was cut off from the buck material, as shown in
Figure 20.
Figure 21 is the side part of the buck material microstructure.
Figure 22 is the machined surface material microstructure. As
Figure 22 shows, there was no surface burn on the workpiece material. The machined surface microstructure grain was smaller than the buck material, showing the machined surface was recrystallized after the dry milling process. The recrystallized surface was thin enough about 185 μm, which could be processed during the semi-finishing or finishing process.
The chip’s form was different between the cutting speed higher than 600 m/min and cutting speed lower than 600 m/min. When the cutting speed was higher than 600 m/min, the chips looked like the power, as shown in
Figure 23. When the cutting speed was lower than 600 m/min, the chips looked like a mixture of long comma chips and short comma chips, as shown in
Figure 24.